me, an internetual

9.7K posts

me, an internetual banner
me, an internetual

me, an internetual

@grrowl

i make cool shit on the internet • • • 🌈 🇦🇺 • • • https://t.co/kbMT4MulaG

Melbourne, AU Katılım Kasım 2007
1.4K Takip Edilen448 Takipçiler
me, an internetual
@KaranVaidya6 > gained a foothold in an internal agentic tool > The attacker demonstrated deep knowledge of our API surface and internal architecture The call is coming from inside the house
English
0
0
0
627
Matthew Lam
Matthew Lam@mattlam_·
i'm sad and confused why claude code would remove /simplify. /simplify is the most useful built in skill, and as it's name, it's simple. Instead now it's going to be /code-review with different effort levels but: - agents are not good enough to write clean code yet, and /simplify still helps majorly - sometimes you want to simplify your code without a full on review - /simplify doesn't need effort level, additional prompt, or comment replies. @bcherny thoughts?
Claude Code Changelog@ClaudeCodeLog

Claude Code 2.1.147 has been released. 35 CLI changes Highlights: • Workflow tool added for deterministic multi-agent orchestration; off by default, set CLAUDE_CODE_WORKFLOWS=1 • /simplify→/code-review renamed; flags correctness bugs at effort level, can post inline GitHub PR comments • REPL and Workflow sandboxes hardened against prototype-pollution and thenable escapes, cutting escape risk Complete details in thread ↓

English
15
1
115
45.2K
Advait Shinde
Advait Shinde@advait3000·
@threepointone cf needs to promote the Agent class to be the default thing over raw DOs. Scheduling, state management, queueing, idempotency, keepAlive are all such great features. 90% of DO impls would benefit from these utilities. It's strange to even call it Agent honestly.
English
5
1
17
15.7K
me, an internetual retweetledi
Pauline P. Narvas
Pauline P. Narvas@paw_lean·
FWIW, I think the best managers are the ones who are the best ICs as well. The rest of the team is forced to level up as a result and the manager never loses touch with the tech/product/community. I never really believed in the role of just "people managers".
English
28
21
237
85.8K
me, an internetual
@the_thagomizer Waiting for “antigravity is now gemini dev” and “Gemini cli is shutting down in favour of google AI”
English
1
0
0
67
Mario Zechner
Mario Zechner@badlogicgames·
people of pi.dev. i'm removing all tools from pi witbout replacement. get creative.
max.berlin@maxjendrall

@YoniBraslaver @badlogicgames oh god, please don't switch out all read, write, edit, bash tools for code mode. Would be in the spirit of extensibility tho hahaha "pi has 1 tool. deal with it"

English
59
6
493
160.3K
me, an internetual retweetledi
International Cyber Digest
International Cyber Digest@IntCyberDigest·
🚨 How the TanStack npm attack actually happened: 1. Attacker opened a normal-looking pull request (#7378) on the TanStack repo. 2. GitHub automatically ran CI tests on that PR. 3. Code inside the PR stole the workflow's GitHub Actions Cache write token during the test run. 4. The attacker used that token to plant poisoned files in the shared build cache. The PR could be closed afterwards. The poisoned cache stays. 5. The official release workflow later pulled from the cache, baked the malicious files into the build, and signed and published 84 malicious package versions to npm.
Adnan Khan@adnanthekhan

This attack leveraged GitHub Actions Cache Poisoning. Payload deployed here: github.com/TanStack/route… It looks like it detonated here: #step:26:2" target="_blank" rel="nofollow noopener">github.com/TanStack/route…

English
61
574
4.7K
806.7K
David Boskovic
David Boskovic@dboskovic·
if your agent doesn't write design specs like this your ngmi
David Boskovic tweet media
English
17
11
354
126.8K
me, an internetual
me, an internetual@grrowl·
@AmpCode bug report — fresh installation crashes poorly when ~/.cache/amp/ doesn't exist.
me, an internetual tweet media
English
0
0
0
56
retto
retto@rettooooo·
@adamghowiba multi-agent is a broken paradigm
English
7
0
12
14.3K
Adam Ghowiba
Adam Ghowiba@adamghowiba·
JP Morgan's investment research team just shared exactly how they built their multi-agent system "Ask David", and it's the same architecture pattern showing up everywhere: - supervisor agent orchestrates - specialized subagents handle retrieval, structured data, analytics - LLM-as-judge reflection node before the answer ships - human-in-the-loop for the last accuracy gap worth watching for anyone building:
English
133
677
7K
2M
Harish Uthayakumar
Harish Uthayakumar@curiousharish·
Someone please build a wrapper on AWS! Most confusing interface ever.
English
249
27
885
98K
Steve Ruiz
Steve Ruiz@steveruizok·
What are you guys using for automated code reviews? Cursor's gracious bugbot trial ran out and Copilot code reviews leave something to be desired
English
35
0
32
11.7K
Alex Kehr
Alex Kehr@alexkehr·
the Lovable app is shockingly bad. how did a team build this and feel proud enough to ship this hyper-slop?
English
225
13
895
153.5K
me, an internetual retweetledi
onionweigher 🧅⚖️
onionweigher 🧅⚖️@onionweigher·
According to 23andme I have that gene that makes cilantro taste like soap, but I love cilantro. Thus I can conclude that disliking cilantro is a moral failing
English
52
163
3.7K
695.9K
sunil pai
sunil pai@threepointone·
now that the basics are falling into place, the thing I'm really struggling with is UI. every "coding agent" kinda sucks on laptop/mobile, and everyone punts on it by just hiding the details "you don't need to see the code anymore". balderdash. we need to rethink it all.
sunil pai@threepointone

working on subagents / agents-as-tools for single threads, and it's so very satisfying with Think because it's just nested chat all the way down, persistence/streaming/resumption all ootb landing this week

English
17
2
64
5.9K